[PATCH weston v7 3/3] drm: Don't hang onto the backend config object post-backend_init
bryce at osg.samsung.com
Wed May 11 01:19:54 UTC 2016
On Tue, May 10, 2016 at 03:33:59PM +0300, Pekka Paalanen wrote:
> On Fri, 29 Apr 2016 15:40:35 -0700
> Bryce Harrington <bryce at osg.samsung.com> wrote:
> > The drm backend was copying most everything out of the config object
> > already, but now also copy the use_current_mode parameter and the
> > config_output function pointer, so that there are no remaining
> > references to the config object passed into backend_init().
> > By decoupling the config struct to the backend, if in the future if the
> > structure definition changes in non-backwards compatible ways, then any
> > version compatibility adaptation will be limited to just the
> > backend_init() routine.
> > With the use_current_mode moved into the main config class, the
> > drm_config wrapper is redundant. Dropping it helps make the drm backend
> > config initialization more consistent with the other backends.
> > Also, enforce destruction of the backend config object after
> > initialization. Since the backend config struct versioning implies that
> > there we expect potential future descrepancy between main's definition
> > of the config object and the backend's, don't allow the backend to hang
> > onto the config object outside the initialization scope.
> > Signed-off-by: Bryce Harrington <bryce at osg.samsung.com>
> > Reviewed-by: Pekka Paalanen <pekka.paalanen at collabora.co.uk>
> > ---
> > v6:
> > - Drop use of drm_config config wrapper
> > v7:
> > - Update to master
> > - Put backend configs on stack instead of zalloc()
> > - Enforce destruction of backend config object
> > (Squashed patch as requested by pq)
> > src/compositor-drm.c | 24 +++++++++++++++---------
> > src/compositor-drm.h | 3 ++-
> > src/main.c | 46 ++++++++++++++++------------------------------
> > 3 files changed, 33 insertions(+), 40 deletions(-)
> I decided to do some cleaning up here, and squashed these two patches.
> It was getting annoying to review one patch and then see the follow-up
> in the same series void some of the earlier comments that would have
> caused a reject.
> To put things short, pushed:
> 59987fa..1c0e40d master -> master
> I also wrote the commit message for this patch from scratch. I do not
> recall any of the conversion patches before this one to actually
> document *why* we are doing this. This and other changes you can find
> listed upstream. I thought I won't bother you with yet another round,
> but just fix what I thought was important at the moment and push it.
> Let's deal with any comments and fallout in follow-ups.
Sounds good, it's nice to see this patch series finally get wrapped up,
and I agree there's plenty more work here for the future.
> Thank you Bryce for picking this up, and Benoit too.
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v2
> -----END PGP SIGNATURE-----
More information about the wayland-devel